计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (20): 9-11.DOI: 10.3778/j.issn.1002-8331.2010.20.003

• 博士论坛 • 上一篇    下一篇

应用多方向光源分割带钢表面缺陷

丛家慧,颜云辉   

  1. 东北大学 机械工程与自动化学院,沈阳 110004
  • 收稿日期:2010-04-02 修回日期:2010-05-24 出版日期:2010-07-11 发布日期:2010-07-11
  • 通讯作者: 丛家慧

Application of varying directed illumination to segment strip surface defects

CONG Jia-hui,YAN Yun-hui   

  1. School of Mechanical Engineering and Automation,Northeastern University,Shenyang 110004,China
  • Received:2010-04-02 Revised:2010-05-24 Online:2010-07-11 Published:2010-07-11
  • Contact: CONG Jia-hui

摘要: 光源的照明方式和安装角度对采集到的图像质量有很大的影响,这种不同会导致同一幅缺陷图像产生不同的纹理效果。目前的采集系统一直将光源的影响看作负面因素,事实上如果将不同方向光源对带钢缺陷图像信息的影响用于分割,能够获得更全面、准确的图像信息,因为光源序列的分割方法比单光源包含了更多的缺陷特性。首先采集不同方向光源照射下的图像序列,再提取图像最大强度值位置对应的光源倾角方向作为特征,最后在特征空间运用线性判别函数得到分割结果。实验结果表明提出的方法能获取更高对比度的缺陷图像,正确地将缺陷从背景图像中分割出来。

关键词: 图像分割, 光源序列, 表面缺陷, 图像对比度

Abstract: The model and installing angle of illumination will affect the captured image quality and then lead to different texture characteristics with the same image.At present,the normal acquisition system regards the influence of illumination as negative factors.But actually this difference between varying illuminations can be used to segment defect images and obtain more complete and correct image information.Because the segmentation algorithm based on illuminant series include more defect features than single illumination.Firstly the image series are obtained in different directed illumination.Then the feature is derived,which corresponds to a tilt angle with maximum image intensity.Finally the segmented result using linear decision criteria is acquired in the feature space.Experimental result shows that this method enables to obtain higher contrast images and segments defects from background accurately.

Key words: image segmentation, illuminant series, surface defects, image contrast

中图分类号: